Regular articleThe neural basis of individual differences in working memory capacity: an fMRI study
Introduction
Working memory refers to a system involved in the temporary storage and processing of information, and it supports higher cognitive brain function such as language comprehension, learning, and reasoning Baddeley 1986, Just and Carpenter 1992. Brain-imaging studies have attempted to identify functional brain anatomy underlying working memory systems based on Baddeley’s theory (Baddeley, 1986). It has been proposed that two types of working memory processes are subserved by distinct cortical structures, the executive control processes, located in the prefrontal cortex, and the modality-specific buffers, located in more posterior regions Awh et al 1996, Paulesu et al 1993, Smith et al 1996. As for executive function, activation of the dorsolateral prefrontal cortex (DLPFC) has been observed when two kinds of tasks were performed together (D’Esposito et al., 1995), when the task was performed with a self-monitoring system (Petrides et al., 1993) or during a task requiring executive control (Cohen et al., 1997).
It has also been suggested that the executive system serves as an attention controller that allocates and coordinates attentional resources for task-relevant stimuli and responses (Baddeley, 1996). Neuroimaging studies have related this attention control system in executive function to activity in frontal regions, particularly the DLPFC and anterior cingulate cortex (ACC) (Smith and Jonides, 1999).
However, there has been little research on individual differences in the neural substrates of working memory capacity.
It has been found that there are individual differences in working memory capacities and that differences in working memory can account for many aspects of language comprehension Just and Carpenter 1992, Baddeley et al 1985. Previously, a reading span test (RST) was developed and implemented to measure, behaviorally, individual differences in the verbal working memory capacity employed by processing and storage functions during reading sentences (Daneman and Carpenter, 1980). Similarly, a listening span test (LST) was also developed to measure working memory capacity during listening to sentences. In the LST, subjects listen to a few sentences and judge whether each sentence is semantically true or not, while maintaining the last word of each sentence.
The working memory resources available for both maintaining and reading or listening are finite, and subjects must allocate portions of these resources during RST or LST. In previous studies, subjects with large working memory capacities (high-span subjects in RST) were successful in maintaining the target words during reading or listening. However, subjects with small working memory capacities (low-span subjects in RST) have difficulty in maintaining target words due to insufficient working memory capacity. Significant high correlations between both RST and LST span scores and reading comprehension scores have been found Daneman and Carpenter 1980, Osaka and Osaka 1994, Osaka 2000. Moreover, there was a significant correlation between RST and LST span scores (Daneman and Carpenter, 1980). These results suggest that both RST and LST essentially measure the same capacity of individual working memory associated with language comprehension, independent of the modality of stimulus presentation (Daneman and Merikle, 1996).
From this perspective, both span tasks would be a powerful predictor of the neural bases in individual differences in verbal working memory capacity.
Moreover, both span tasks are excellent for testing parallel control functions in executive systems of working memory, because sentence comprehension demands extensive storage of partial and final products in the service of complex information processing, as well as requiring maximum attentional control (Engle et al., 1999).
Using fMRI, the neural substrates of span tasks have been investigated and increases in activation associated with task demands were observed in left frontal and temporal language areas (Just et al., 1996a) and in prefrontal cortex (PFC) (Bunge et al., 2000). However, individual differences in the recruitment of such neural substrates devoted to verbal working memory capacity during the span task have not yet been explored.
Using MEG to investigate neural activity during LST, Osaka et al. (1999) found activation differences between two subjects groups divided by RST span scores.
Following this evidence, further questions arise concerning the neural bases attributable to the differences between high-span subjects and low-span subjects: What are the neural substrates of working memory resources and how do these relate to individual performance difference? Which of these brain mechanisms are required to perform the span task?
To answer these questions, in the present experiment we compared the neural substrates, which are attributable to differences between the high-span and the low-span subjects’ performance on a span task. We selected two groups of subjects; high-span subjects and low-span subjects according to the span scores on the RST. Then, we used fMRI to measure brain activity associated with the performance of LST and compared fMRI activations between high-span subjects and low-span subjects. We employed three experimental conditions: LST, Listen, and Remember conditions. The LST condition was a dual-task paradigm, in which subjects were required to both listen to each sentence and remember the target words. The Remember and Listen conditions were single-task paradigms of maintaining the target words and listening to sentences, respectively.
Section snippets
Subjects
The subjects were college students or graduates aged 20–27. Two groups of nine high-span subjects (HSS) who had span scores of RST ranging from 4.0 to 5.0 and nine low-span subjects (LSS) who had span scores ranging from 2.0 to 2.5 each based on their RST scores Osaka and Osaka 1992, Osaka and Osaka 1994 were selected. The span value was evaluated as the highest level at which subjects could correctly recall the target word of each sentence; e.g., if they recalled all the target words of four
Results
Two subjects, one in each of the HSS and LSS groups were eliminated from the analysis because of their excessive head movement. Following analysis was done for eight subjects in each of the HSS and LSS groups.
Discussion
The present fMRI study showed that main activation areas appeared in three regions while the subjects were engaged in the LST: temporal, PFC, and ACC areas. These results suggest that the neural substrates of verbal working memory involve interconnections among these areas.
The first region of the network system is around the Sylvian fissure; particularly, the superior temporal gyrus near Wernicke’s area. Just et al. (1996b) found an increase of activation in these areas when the sentence
Acknowledgements
The work was supported by a Grant-in-Aid from the Japan Society for the Promotion of Science (14310041) to M.O., (12301005) to N.O., and a Grant-in-Aid for Research for the Future Program JSPS-RFTF97L00201 from the Japan Society for the Promotion of Science to H.S.
References (37)
The episodic buffera new component of working memory
Trends Cogn. Sci.
(2000)- et al.
Components of fluent reading
J. Memory Lang.
(1985) - et al.
A parametric study of prefrontal cortex involvement in human working memory
NeuroImage
(1997) - et al.
Cognitive and emotional influences in anterior cingulated cortex
Trends Cogn. Sci.
(2000) - et al.
Individual differences in working memory and reading
J. Verb. Learn. Verb. Behav.
(1980) - et al.
Collaborative activity between parietal and dorso-lateral prefrontal cortex in dynamic spatial working memory revealed by fMRI
NeuroImage
(2000) - et al.
Effect of focus on verbal working memorycritical role of the focus word in reading
Memory Cogn.
(2002) - et al.
Dissociating verbal and spatial working memory using PET
Cereb. Cortex
(1996) - et al.
Dissociation of storage and rehearsal in verbal working memoryevidence from PET
Psychol. Sci.
(1996) Working memory
(1986)
Working memory
Science
Exploring the central executive
Q. J. Exp. Psychol.
Anterior cingulated cortex and response conflicteffects of frequency, inhibition and errors
Cereb. Cortex
A resource model of the neural basis of executive working memory
Proc. Natl. Acad. Sci. USA
The counting stroopan interference task specialized for functional neuroimaging: validation study with functional MRI
Hum. Brain Mapp.
Temporal dynamics of brain activation during a working memory task
Nature
The magical number 4 in short-term memorya reconsideration of mental storage capacity
Behav. Brain Sci.
Working memory and language comprehensiona meta-analysis
Psychonom. Bull. Rev.
Cited by (151)
Impacts of visuo-spatial working memory on the dynamic performance and safety of car-following behavior
2024, Physica A: Statistical Mechanics and its ApplicationsThe content-dependent effect of the N-back task on dual-task performance
2023, Behavioural Brain ResearchThe role of neural load effects in predicting individual differences in working memory function
2021, NeuroImageCitation Excerpt :As discussed above, it is only more recently that fMRI studies have provided sample sizes necessary for conducting a comprehensive analysis of between-subjects WM variation. Thus, most of the prior published work on this topic either had exceedingly low power for replication (Bunge, 2001; Todd and Marois, 2005), or instead focused on isolated a priori regions of interest (e.g., Osaka et al. 2003). By taking advantage of the large sample size of the HCP and a whole-brain parcellation scheme, the current study provides a powerful approach by which to resolve prior discrepancies in the literature.
Quantifying the variability of neural activation in working memory: A functional probabilistic atlas
2021, NeuroImageCitation Excerpt :The results showed that working memory span varied from 2 to 5 words across individuals (Daneman and Carpenter, 1980). To reveal the underlying neural mechanism in brain activity, fMRI studies have compared the high- and low-memory span groups and found high span group had significantly stronger activation in the prefrontal regions, inferior frontal gyrus, and anterior cingulate cortex (ACC) (Callicott et al., 1999; Jansma et al., 2004; Kharitonova et al., 2015; Mattay et al., 2006; Newman et al., 2013; M. Osaka et al., 2003; N. Osaka et al., 2004). More recently, a study reported that the high-capacity group also showed greater activation in parietal areas (Liu et al., 2018).